- Joined
- Oct 10, 2018
- Messages
- 4
- SL Rez
- 2007
I travel quite frequently. I don't sell anything and most of my creations are for personal use. I've used SIMONASTICK for years when traveling. However as OpenSimulator has added features I've gotten hooked on variregions. I've tried almost all of the grids and have used 4x4, 6x6, 8x8, 10x10 and currently using a massive 25x25. The problem is two-fold. As you expand so does lag.
I'm a hardware guy and understand it pretty well. Simonastick is about the maximum level of my software competence. What I want to build is an offline single region that is either another 25x25 of the max for OpenSimulator, which is 32x32. This allows me to take Google Maps and create massive landscapes based on real world terrain. My current project running on one of the professional grids is a replica of North Carolina's Pisgah National Forest. However I'm seeing LAG based on people using the grid.
What I want to do is eliminate people as an issue (and their assets, if I'm understanding the dependencies here).
I've read that simply loading a massive region the problem becomes memory. So, my questions are:
Can I reduce lag by simply adding more memory to a standalone server? Would more cores help? Would replacing the normal Intel I series with multiple Xeon cores give me any advantage or adding more GPUs (I currently use dual Nvidias in SLI mode, which I know is not much help client side. Would there be an advantage in using professional level GPUs similar to the Nvidia Quadro? Like I said, I know the hardware but don't understand the software well enough to know what I need - I'm sort of unconsciously incompetent - I don't even know what I don't know.
The idea system should be primarily standalone, occasionally be able to attach to OSGrid or Metropolis and use Hypergrid. It should support no more than 5 avatars at max load and I would probably limit it to 500k prims since I am primarily using this for scenic walkaround projects. Scripts would be limited to basic scripts for waterfalls, some doors (both pretty much rotation based) and other scenery related things and the usual avatar AO attachments.
I would prefer to build my own hardware, but would not be adverse to using one of Amazon's AWS solutions for on/off and would like to keep the costs below about $300 USD if using a cloud based solution. Lots of stuff here and as I said not even sure of the right questions to ask. When I've asked in several grids the answers have been pretty evasive almost like they think I'm looking to compete rather than use this for some of my mobility limited friends, who can no longer get to these locations.
Thanks for any help
I'm a hardware guy and understand it pretty well. Simonastick is about the maximum level of my software competence. What I want to build is an offline single region that is either another 25x25 of the max for OpenSimulator, which is 32x32. This allows me to take Google Maps and create massive landscapes based on real world terrain. My current project running on one of the professional grids is a replica of North Carolina's Pisgah National Forest. However I'm seeing LAG based on people using the grid.
What I want to do is eliminate people as an issue (and their assets, if I'm understanding the dependencies here).
I've read that simply loading a massive region the problem becomes memory. So, my questions are:
Can I reduce lag by simply adding more memory to a standalone server? Would more cores help? Would replacing the normal Intel I series with multiple Xeon cores give me any advantage or adding more GPUs (I currently use dual Nvidias in SLI mode, which I know is not much help client side. Would there be an advantage in using professional level GPUs similar to the Nvidia Quadro? Like I said, I know the hardware but don't understand the software well enough to know what I need - I'm sort of unconsciously incompetent - I don't even know what I don't know.
The idea system should be primarily standalone, occasionally be able to attach to OSGrid or Metropolis and use Hypergrid. It should support no more than 5 avatars at max load and I would probably limit it to 500k prims since I am primarily using this for scenic walkaround projects. Scripts would be limited to basic scripts for waterfalls, some doors (both pretty much rotation based) and other scenery related things and the usual avatar AO attachments.
I would prefer to build my own hardware, but would not be adverse to using one of Amazon's AWS solutions for on/off and would like to keep the costs below about $300 USD if using a cloud based solution. Lots of stuff here and as I said not even sure of the right questions to ask. When I've asked in several grids the answers have been pretty evasive almost like they think I'm looking to compete rather than use this for some of my mobility limited friends, who can no longer get to these locations.
Thanks for any help